topic How to properly display trend comparison of values over Hour by Day for multiple days? in Splunk Search

How to properly display trend comparison of values over Hour by Day for multiple days?

ten_yard_fight — Tue, 17 Dec 2019 00:52:55 GMT

The following query will display a simple chart for trend comparison. This works well if you keep the days you're comparing =< 4 days.
The Query:

index="my_index"
| search src="all_sources" 
| bin _time as Time
| eval Day=strftime(Time,"%m-%d")
| eval Hour=strftime(Time,"%H") 
| chart eval(round(avg(transfer_duration), 2)) as AvgXferDuration over Hour by Day

The above query will produce the following graph. (as long as its =< 4 days)

Now, if you increase the days to > 4 days. For example, if you want to see a week's trend (7 days). You will get the following graph. Notice the graph lines become scattered points in the center. (not sure why this happens)

What is the proper way to display the graph for multiple days? e.g. > 4 days, 7 days, etc.

Re: How to properly display trend comparison of values over Hour by Day for multiple days?

to4kawa — Tue, 17 Dec 2019 12:48:54 GMT

| makeresults count=2 
| streamstats count 
| eval _time = if (count==2,relative_time(_time,"-2month@d"), relative_time(_time,"@d")) 
| makecontinuous _time span=15min 
| eval transfer_duration=random() % 3.50 + .5 

| rename COMMENT AS "Everything above generates sample event data; everything below is your solution." 

| eval Day=strftime(_time,"%m-%d") 
| eval Hour=strftime(_time,"%H") 

| rename COMMENT AS "this is key logic" 

| where _time >relative_time(now(),"-7d@d") 

| chart eval(round(avg(transfer_duration), 2)) as AvgXferDuration over Hour by Day

Re: How to properly display trend comparison of values over Hour by Day for multiple days?

ten_yard_fight — Tue, 17 Dec 2019 18:36:01 GMT

Thank you very much @to4kawa . I see where my issue was. Your example helped me pay attention to the details.
Turns out if I DO NOT bin _time, but instead remove it from my query, the chart command processes my transform correctly.

Again, thank you for your suggestion. Please enjoy some of my karma.